HAL Id: hal-02991093https://hal.archives-ouvertes.fr/hal-02991093

Submitted on 4 Sep 2021

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

Logophoricity and shifts of perspectiveTatiana Nikitina

To cite this version:Tatiana Nikitina. Logophoricity and shifts of perspective. Functions of Language, John BenjaminsPublishing, 2020, 27 (1), pp.78-99. �10.1075/fol.20001.nik�. �hal-02991093�

1

To appear in Functions of Language 27(1): 78-99

https://doi.org/10.1075/fol.20001.nik

Logophoricity and shifts of perspective:

New facts and a new account*

Tatiana Nikitina (CNRS-LLACAN)

This study presents a typology of existing approaches to logophoricity and discusses problems the different

approaches face. It addresses, in particular, perspective-based accounts describing constructions with

logophoric pronouns in terms of their intermediate position on the direct-indirect continuum (Evans 2013),

and lexical accounts incorporating the idea of coreference with the reported speaker into the pronoun’s

meaning, either through role-to-value mapping mechanisms (Nikitina 2012a,b), or through feature

specification (Schlenker 2003a,b). The perspective-based approach is shown to be unsatisfactory when it

comes to treating language-specific data in precise and cross-linguistically comparable terms. It fails to

account, for example, for cross-linguistic differences in the behavior of logophoric pronouns, for their

optionality, and for their close diachronic relationship to third person elements. Lexical accounts are better

equipped to handle a variety of outstanding issues, but they, too, need to be revised to accommodate a

variety of discourse phenomena associated with logophoricity, including alternation with first person

pronouns. The proposed solution follows the lines of lexical approaches but aims at enriching the pronouns’

lexical representation with notions pertaining to narrative structure, such as the role of Narrator. A separate

solution is proposed for treating conventionalized uses occurring outside speech and attitude reports.

* This project has received funding from the European Research Council (ERC) under the

European Union’s Horizon 2020 research and innovation programme (grant agreement No

758232).

2

1. Introduction

1. 1 The puzzle of logophoricity

An area where the notion of perspective has been used most prominently is the study of reported

discourse, since perspective is commonly assumed to underlie the distinction between indirect

and direct speech (Coulmas 1986: 2). Despite remaining largely pre-theoretical, the notion of

perspective is widely applied to newly described strategies, no matter how dissimilar to canonical

instances of direct and indirect speech they may look: strategies not conforming to European

direct and indirect speech tend to be described as mere deviations from the “direct” and the

“indirect” ideals (Evans 2013; Güldemann & von Roncador 2002; von Roncador 1988, inter alia).

Among such deviating strategies logophoric speech is one of typologists’ favorites: it has

been described repeatedly in terms of mixed perspective, and characterized as “semi-direct”,

“semi-indirect”, “combined”, “neutralized”, and “biperspectival” (Aikhenvald 2008; Boyeldieu

2004b; Evans 2013; Thomas 1978). This study is a critical assessment of the usefulness of

perspective-based approaches to the phenomenon of logophoricity, with implications for the study

of reported speech more generally.

I review three major recent theoretical proposals regarding the nature of logophoricity,

two lexical (Nikitina 2012a; Schlenker 2003a,b) and one essentially constructional1 (Evans 2013).

I show that all three fail to account for the way logophoricity functions. My proposed tentative

solution blends the notion of perspective, widely used in the study of direct/indirect speech, with

a new structural dimension: narrative roles, which differ both from semantic roles and from

speech act participant roles.

1 My use of the term “construction” is more restrictive than the use adopted in the Construction

Grammar tradition: I only use it to refer to fixed form-meaning pairings at the level above the

word. Nothing hinges on this in my argument, however.

3

1.2 Three recent accounts: a contrastive analysis

Logophoric pronouns appear in two types of context that traditionally receive different treatment

in formal semantics: in reported speech, they refer to the speaker of the reported speech event,

and in attitude reports, they refer to the participant to whom the attitude report is attributed

(Clements 1975; Hagège 1974) 2. Logophoricity is characteristic of West and Central African

languages, but it is by no means restricted to Africa (cf. Bugaeva 2008; Nikitina & Bugaeva fc.

on Ainu).3 A typical example is presented below: the participant coded by the logophoric

pronouns in (1a) is co-referential with the reported speaker.

(1) Wan (Nikitina 2012b):

a. ɓe a nɔ ge ɓa ɓe gomɔ

then 3SG wife said LOG that.one understood

‘And his wifei said shei had understood that.’

b. ɓe a nɔ ge e ɓe gomɔ

then 3SG wife said 3SG that.one understood

‘And his wife said he had understood that.’

2 The phenomenon of logophoricity cuts across the distinction between the reporting of speech

and the reporting of cognitive relation to a proposition, motivating researchers to treat discourse

reports as a type of attitude report and to assign, erroneously, properties of indirect discourse to

logophoric reports of lesser-studied languages; I return to this issue below.

3 I am concerned here primarily with specialized logophoric pronouns that do not appear outside

logophoric contexts. Many languages lack a dedicated logophoric pronoun but employ reflexive

and other pronouns in a logophoric function; I return to this type of language below, in Section

2.3.

4

Three different proposals have been made recently regarding the meaning of logophoric

pronouns.

Schlenker (2003a) treats logophoric pronouns as shifted indexicals with unusual scopal

behavior. In English, a first person pronoun must be evaluated with respect to the actual speech

act. In some languages first person pronouns can be evaluated with respect to the context of a

reported speech act (Amharic), and there are also languages that use pronouns that must be

evaluated with respect to the context of a reported speech act (logophoric languages). On this

account, logophoric pronouns are specified as denoting a speaker of some speech act (+author),

but they are not allowed to refer to the speaker of the actual speech act (–actual). Constructions

with logophoric clauses are treated as “standard cases of indirect discourse without any

quotational intrusion” (Schlenker 2003a: 411), and their difference from English indirect speech

derives essentially from the pronoun’s lexical specification.

Nikitina (2012a) presents another version of a lexical account, based on a system of

mapping of participant roles onto pronominal values. Languages are assumed to vary in the way

deictic distinctions are drawn within the pronominal system. In English, pronominal deixis is

based on the opposition between actual speech act participants (mapped onto first and second

person) and all others. Some languages may treat participants of reported speech acts in the same

way as actual speakers and addressees (e.g., Havyaka Kannada; Bhat 2007: 59-61), others may

neutralize that distinction for addressees but not for speakers (Adioukrou; Hill 1995), and there

are also languages that employ dedicated pronouns to refer to participants of reported speech acts

only (the logophoric languages). Nikitina’s account is similar to Schlenker’s in exploring cross-

linguistic differences in pronominal meaning, but it is free from the assumption that logophoric

pronouns are associated with indirect discourse. It predicts, accordingly, that across languages,

logophoric pronouns may function differently with respect to the direct/indirect distinction, in

particular – with respect to the way other pronouns are assigned their values.

In sum, the lexical accounts assume that pronouns have different meanings in different

languages. Coached in the formal semantic tradition, Schlenker’s account treats logophoricity in

5

the context of attitude reports, effectively assuming that logophoric pronouns appear in indirect

discourse. Nikitina’s account makes no such assumption, predicting instead that the direct/indirect

distinction should be orthogonal to the use of logophoric pronouns, and languages should be found

where logophoric pronouns appear in what is essentially direct speech.

Evans (2013) explores a rather different line of argument (see also Güldemann & von

Roncador 2002; von Roncador 1988). The cross-linguistic variation is located not in differences

in pronominal meaning but rather in the behavior of language-specific reported speech

constructions. Following the tenets of canonical typology, structures recruited to represent

reported discourse are placed on a continuum between the ideal types of European-style direct

and indirect speech. The ideal constructional types are, furthermore, defined in terms of

perspective. Logophoric pronouns are described as a special case involving a “double deictic

perspective” (Evans 2013: 89). Their value is calculated simultaneously from the perspective of

two different speech acts: they refer “to a person who was the speaker in the reported speech

event, but is third person in the primary speech event” (Evans 2013: 90).

Table 1 summarizes the basic properties of the three accounts. Below I argue that all three

fail to handle the full set of uses of logophoric pronouns, even in a single language. In Section 2,

I discuss issues that only pose problems for some of them; in Section 3, I turn to issues that all of

them leave unexplained. Tentative solutions are presented in Section 4, and their implications

discussed in Section 5.

6

Table 1. Three major approaches to logophoricity

type of account meaning of logophoric

pronouns

logophoric contexts

lexical feature-based

(Schlenker 2003a,b)

author of a reported speech

act

indirect discourse (attitude reports)

lexical role-based

(Nikitina 2012a,b)

reported speaker speech reports, direct or not

construction-oriented

(Evans 2013)

double perspective speech reports intermediate between

direct and indirect discourse

2. Little-discussed properties of logophoric systems

2.1 Logophoric pronouns are not restricted to indirect discourse

A common misconception regarding the use of logophoric pronouns has to do with their relation

to indirect speech; it derives from the formal semantic tradition that subsumes reported speech

under propositional attitude reports. According to Schlenker (2003a: 423), logophoric pronouns

“appear solely in indirect discourse”; according to Sells (1987: 475), they “really are, then,

pronouns that occur in contexts of indirect discourse”. That logophoric pronouns appear in

indirect discourse is also taken for granted in Culy (1997), Frajzyngier (1985), Hyman & Comrie

(1981), among many others.

While the commonly cited examples are largely compatible with this claim, a closer look

at a wider range of data reveals that logophoric pronouns combine with all sorts of elements that

qualify the report as “direct speech” (assuming, of course, that the direct/indirect distinction is

universally relevant, as so many authors do). Nikitina (2012a) discusses examples from different

languages showing that logophoric pronouns freely combine with interjections, vocatives,

7

imperatives and questions, as well as with unambiguous instances of quotation, as in example (2)

(see also Perrin 1974; von Roncador 1988, 1992).

(2) Goemai (Hellwig 2006):

ji t’al oelem

SG.M.LOG.SP pluck “beans”

‘(He said that) heLOG plucked the beans’ (the childish form oelem is used instead of the

standard adult oerem ‘beans’)

There is little evidence, beyond pronominal deixis itself, for treating reports involving

logophoric pronouns as instances of indirect discourse, even if such a notion applied to a

logophoric language (Goddard & Wierzbicka 2019; Nikitina 2012b; Nikitina & Bugaeva fc.;

Nikitina & Vydrina 2020).4

The fact that logophoric pronouns are not restricted to indirect discourse poses problems

to Schlenker’s account, as it undermines the comparison between English first person (which is

never shifted in indirect speech) and logophoric pronouns. It is, on the other hand, fully

compatible with Nikitina’s account (which treats the direct/indirect distinction as orthogonal to

questions of pronominal meaning) and Evans’ account (which builds on the idea that languages

– including logophoric languages – deviate in various ways from the ideals of “direct” and

“indirect” speech).5

4 It may seem paradoxical that reported speech that includes a logophoric pronoun need not be

“indirect”: after all, the logophoric pronoun clearly could not be part of the original discourse.

One should not be led to believe, however, that a “direct” report can be taken as a literal

representation of some sort of original discourse: no report can reproduce all aspects of the

original utterance, and which aspects are chosen as relevant enough to be reproduced is to a

large extent a matter of convention (Plank 1986).

5 All three accounts, as we will see below, have little to say about the fact that logophoric

pronouns occur outside reported speech.

8

2.2 Logophoricity comes in varieties

Theoretical treatments of logophoricity commonly ignore differences among logophoric

languages. Consider the logophoric clauses in (3) and (4), from Ewe and Wan, respectively. Both

reports feature a dedicated logophoric pronoun referring to the speaker of the reported speech act.

Yet the addressee of the same speech act is treated differently: it is encoded by a third person

pronoun in Ewe but by a second person pronoun in Wan.

(3) Ewe (Clements 1975):

Kofi gblɔ na wo be ye-a-dyi ga-a na wo

K. speak to 3PL that LOG-T-seek money-D for 3PL

‘Kofi said to them that heLOG would seek the money for them.’

(4) Wan (Nikitina 2012b: 290):

e ge zo ɓe la ɓa poli

3SG say come then 2SG LOG wash

‘She said: come and wash me.’ (literally, “She said: come and you wash meLOG.’)

The difference in the treatment of addressees suggests, first of all, that there is variation

among logophoric languages with respect to the structuring of the entire pronominal system.

Second, it confirms the idea discussed in the previous subsection: even in terms of person deixis,

logophoric reports cannot be treated as universally “indirect”.

The same language may show a mixture of “direct” and “indirect” values, so that different

pronouns are evaluated with respect to different contexts. In (5), from Wan, a logophoric pronoun

combines with a first person plural pronoun, interpreted with respect to the actual speech act. This

interpretation is in contrast with (4) where a second person pronoun was evaluated with respect

to the reported speech act.

(5) ɓe ge ɓaa ka tɔgɔle do tɛ-ŋ

that.one say LOG+COP 1+3 elder.brother one kill-PROSP

9

‘He wanted to kill one of our elder brothers.’ (literally, “He said heLOG was going to kill

an elder brother of ours.”)

In Wan, in the context of a logophoric pronoun, all pronouns other than first person singular –

including first person dual, first person plural, second person singular, and second person plural

– can be evaluated either with respect to the actual speech act or with respect to the reported

speech act.

How can the difference between Ewe and Wan be accounted for? It is hard to see how it

could be captured on Evans’ account. Both languages would be described as employing

“biperspectival” (= logophoric) pronouns, but they would presumably be placed on different

points of the direct-indirect continuum. This would not be enough, however, to capture the fact

that in Wan, the same pronoun could be evaluated from two different points of view, depending

on broader context.

The same report, moreover, can involve pronouns with conflicting types of evaluation, or

an irregular perspective shift (Gentens et al. 2019; Spronck et al. this issue). In (6), one of the

pronouns – the plural ‘you’ – is evaluated from the perspective of the current speech act, another

– the singular ‘you’ – from the perspective of the reported speech act, and one more – the

logophoric pronoun – from two different perspectives at the same time.6 The perspective-based

account does not provide us with a way of dealing with this partially constrained variation of

interpretation or predicting how particular person values would be interpreted.

(6) ke tuabu e pe ge a la peɓɔ gla gɛ o

if European DEF said say 2PL 2SG baggage take PRT PRT

mɔ ɔ la peɓɔ gla

6 Example (6) involves an embedded speech report with an omitted matrix clause (“They said”),

and the two perspectives are, strictly speaking, associated with a reported speech act and with an

embedded reported speech act. The same combination of person values is attested in non-

embedded speech reports.

10

LOG.PL+COP 2SG baggage take

‘[And they said:] If the white people say that we should carry you, we will carry you.’

Literally: ‘If the white people say let you (pl) carry you, weLOG will carry you.’

On the other hand, by placing the speech reporting strategies of Ewe and Wan on two

different points of the same direct-indirect continuum, the perspective-based account would

presumably predict subtle differences between the two strategies in terms of perspective or

perhaps in the degree of commitment to the reported proposition. Yet there is no discernible

difference in this aspect of the use of logophoric reports in the two languages.

Unlike the construction-based account, both lexical accounts are well equipped to handle

this type of data. On Schlenker’s (2003a) featural account, some pronouns could be assumed to

be underspecified in Wan for the context in which they are evaluated. This would allow second

person pronouns, for example, to be interpreted freely in the context of the actual or the reported

speech act. No such flexibility would be allowed in Ewe, hence third person pronouns would only

be used to refer to participants of the reported speech act. On Nikitina’s (2012a,b) role-to-value

mapping account, Ewe and Wan instantiate different strategies for associating pronouns with

speech act participants, and variation of this kind is in fact predicted to exist. Ewe restricts first

and second persons to actual speech act participants, and in Wan, both actual speech act

participants and participants of a reported speech act are mapped onto second person values (7):

(7) Mapping of participant roles onto pronominal values in Ewe and Wan:

a. actual_sp actual_addr reported_sp reported_addr others (Ewe)

1st 2nd log 3rd

b. actual_sp actual_addr reported_sp reported_addr others (Wan)

1st 2nd log 2nd 3

2.3 Logophoricity need not involve special pronouns

11

Another issue that is commonly misrepresented in the literature has to do with the definition of

logophoric languages. It is generally acknowledged that a logophoric language distinguishes the

actual speaker from the participant to which reported speech or an attitude report is attributed.

While some languages rely on dedicated logophoric pronouns, many more use other means to

achieve the same goal: in particular, many languages use reflexive or third person pronouns in a

logophoric function. Culy (1994) describes the two types as languages with pure and mixed

logophoricity, respectively. The two language types, however, are in reality very closely related,

since they can only be distinguished based on contexts in which elements with the logophoric

function occur: languages with mixed logophoricity differ from pure logophoric languages in that

they do not restrict the use of “co-reference” pronouns to attitude reports (if they did, the pronouns

would be described as logophoric rather than e.g. reflexive).

What is rarely recognized is the gradient nature of the distinction. In a number of

languages, dedicated logophoric pronouns are lacking, but first person pronouns still cannot refer

to the speaker of a reported speech act, unless that speaker is at the same time the actual speaker.

Instead of logophoric pronouns, third person pronouns are used in such cases.

Crucially, languages of this type share important properties with logophoric languages

(Nikitina 2012a), including characteristically “direct” features of the reports (the deixis of direct

speech, interjections, unambiguous elements of quotation) and unusual combinations of person

values. In (8), for example, third person pronouns are not restricted to speech and attitude reports,

yet such reports represent a strategy clearly distinct from both direct and indirect speech, and

identical in many respects to the logophoric strategy.

(8) Obolo (Aaron 1992: 232)

ogwu uga okekito ito ikibe gwun kan , ɔmɔ ikatumu

this mother was.crying cry say child 3SG 3SG not.told

inyi owu ye ibe owu kagɔɔk ifit ifit yi

give 2SG Q say 2SG not.follow play play this

12

‘The mother was crying, saying: My child, did I not tell you not to join in this dance

group?’ (Literally: “The mother was crying, saying: Her child, did she not tell you...”)

Such patterns of “latent” logophoricity are found across West Africa, but rarely

considered together with overt logophoricity. Latent logophoricity is a first step towards the

development of a fully-fledged logophoric system: as old third person pronouns are replaced by

new ones, they may become restricted to speech reports, evolving into dedicated logophoric

markers. This explains the fact that third person elements are a common source of logophoric

markers (Dimmendaal 2001; Hyman 1979; Hyman & Comrie 1981; von Roncador 1992, pace

Clements 1975; Faltz 1985). Dedicated logophoric markers develop in several stages from the

strategy of using third person elements in the logophoric function, and an adequate account of

logophoricity should be capable of describing this type of change.

Lexical accounts can handle the gradient nature of logophoricity relatively easily. The

change from latent to overt logophoricity can be represented as a change in the lexical

specification of the relevant pronouns. On a featural account, the old third person pronoun or a

demonstrative could be assumed to change its meaning to [+reported author], restricting its

reference to authors of reported speech acts. It would continue to refer to authors of attitude

reports, but it would no longer refer to non-speech-act-participants. Such an account presumes

that features such as [+/- author] are basic enough to replace other pronominal meanings, even

though the account offers no principled explanation for this type of change.

On the role-to-value mapping account, what used to be a common non-speech-act-

participant pronoun (third person) would be replaced by a new form, only retaining the role of

reported speaker (9)-(10). This again presupposes that the roles such as “reported speaker” and

“reported addressee” are universally available primitives of pronominal deixis.7

(9) Latent logophoricity (Obolo)

7 One promising approach to treating this type of diachronic change on a role-to-value mapping

account is presented by amphichronic semantic maps (Nikitina 2019, Nikitina & Treis fc.).

13

actual_sp actual_addr reported_addr reported_sp others

1st 2nd 3

(10) Overt logophoricity (Wan)

actual_sp actual_addr reported_addr reported_sp others

1st 2nd log (< *3) 3

The perspective-based approach, on the other hand, does not provide a ready account of

the diachronic facts. When third person pronouns are used in a logophoric function, as in (8), they

would probably qualify the construction as semi-direct or indirect, depending on the way other

pronominal values are assigned; the Obolo construction, for example, is described by Aaron

(1992: 232) as “semi-indirect”. To develop into a dedicated logophoric pronoun, the old third

person pronoun would have to be reanalyzed as a biperspectival marker. Since the current version

of the account operates in terms of entire constructions, it offers no mechanism for locating

changes in specific lexical items. It is therefore unclear, on the current version of the account,

how and why the relevant change in the use of specific pronouns would take place.

Overall, the facts discussed in this section are better compatible with a lexicalist account,

and favor the role-to-pronoun mapping, as it does not rely on the distinction between direct and

indirect speech. That distinction, as we already saw, is orthogonal to the use of logophoric

pronouns. Even the role-to-pronoun mapping account, however, faces challenges when discourse

data are explored in more detail.

3. Issues problematic for all three approaches

3.1 Logophoric pronouns are not restricted to attitude reports

14

One major issue rarely addressed in the literature is the need to better define the range of contexts

beyond reported speech where logophoric pronouns appear. Typical contexts in which they are

found range from reports of thought and emotion to reports of direct perception (Culy 1994;

Stirling 1993).8 They have also been attested in contexts that cannot be easily subsumed under

the notion of attitude report (Frajzynger 1993, inter alia). In Wan, for example, logophoric

pronouns occur in reports of mental and psychological states, in descriptions of purpose, intention

(11a) and, less trivially, point of view (11b). In (11b), the logophoric pronoun is used to compare

the distance as it appeared to a character in the story to the distance described for the actual speech

situation: the distance covered by a child entering the water was the same as the distance from the

actual speaker to Seyi’s house, at the moment of speaking.9

(11) a. ɓe e kuna ta ɓa ɔ e gla

then 3SG ascend on LOG salt DEF take

‘And he climbed up in order to take the salt...’

b. e wia yi e go taiii a ga e

3SG entered water DEF in until 3SG go DEF

ɓa ɓo lapea Sɛyi muu ku e gɔŋ kege gɛ

LOG reached like S. PL+POSS house DEF like like.this PRT

‘He entered the water and went until he was like [from here] to the house of

Seyi’s family.’

In (11a,b), the logophoric context is not introduced by any particular verb, and in fact, no

specific verb could be added to these examples to signal unambiguously that they involve attitude

reporting. In this sense, logophoricity has no lexical licensor, overt or implicit.

8 As discussed in Spronck & Nikitina (2019), defining the notion of speech report in a cross-

linguistically consistent way is itself a non-trivial task.

9 The indexical nature of the distance construction is confirmed by the use of a manner

demonstrative ‘like this’, see Nikitina & Treis (fc.).

15

Even more problematically, logophoric pronouns occur in contexts that cannot be

possibly constructed as involving any attitude report at all, such as in aspectual constructions

(Nikitina 2018a). In (12)-(13), a construction involving the verb ‘say’ is used to render an

aspectual meaning (prospective).

(12) yi e ge ɓa kɔ

water DEF say LOG boil

‘The water was about to boil.’ (lit., ‘The water said: let me boil!’)

(13) ke wati ge ɓa ɓo mɔ

if time say LOG arrive PRT

‘When the time comes...’ (or ‘When the time is about to arrive.’)10

Examples of this sort present a problem to all existing accounts of logophoricity, which

are inherently synchronic and do not capture diachronic meaning extensions. Examples (12)-(13)

cannot be explained in terms of a [+author] feature, and they do not involve the role of reported

speaker. There is also nothing in their interpretation suggesting a double perspective, as the

entities involved can be hardly construed as having their own perspective (these are regular

aspectual constructions involving no personification).

To a construction-based approach, examples of this sort present a challenge of form-

function mismatch. They behave as perspective persistent constructions in the sense of Gentens

et al. (2019) and Spronck et al. (this issue): while logophoric reports are normally expected to be

associated with a double perspective, the aspectual constructions show no evidence for any

perspective shift.

10 In (12)-(13) the logophoric pronoun is in the subject position, suggesting that it could come to

be reanalyzed, in time, as a complementizer in some types of clause. This development,

however, is likely hindered by the pronoun’s frequent occurrence in non-initial positions, as

well as by a general restriction on subject omission (Wan is a non-pro-drop language).

16

3.2 Logophoric pronouns can refer to the actual speaker

Both lexical accounts postulate a fundamental meaning difference between first person and

logophoric pronouns, predicting that they should appear in non-overlapping sets of contexts.

According to Schlenker, “there may not exist any logophoric pronouns that denote the author of

the actual speech act” (2003a: 423). Nikitina (2012a), too, assumes participant roles such as

“speaker” and “addressee” as primitive universal notions. Yet in reality, the notion of person is

no less intuitive and perhaps not much better defined than the notion of perspective.

Since at least Benveniste (1956), the meaning of pronouns has been a subject of

ethnographic research exploring how categories of participant roles vary across languages and

contexts, and suggesting that accounts building on such simplistic notions as “speaker” and

“addressee” fail to do justice to the ways first and second person are used in real discourse

(Goffman 1981; Levinson 1988). First person pronouns do not always refer to the actual speaker

(Urban 1989; Rumsey 2000), their use depending on social construction of personhood and genre-

specific sets of prescribed speech roles (Hanks 1996). Actual speakers are not always represented

using first person pronouns, as evidenced, for example, by the use of third-person titles and proper

names to refer to speech act participants.11

This observation has ramifications for accounts of logophoricity. On the one hand,

speakers who are not participants in the actual speech act can be referred to using first person

pronouns, rather than logophoric pronouns; this will be shown in the next subsection. On the other

hand, actual speakers can be represented by second and third person, and logophoric pronouns

are then used to report their speech.

11 Wan speakers often use proper names, in formal settings and when speaking French, to refer

both to themselves and to their addressees. Speakers of English avoid using first and second

person pronouns in certain types of speech situation, including rituals and speech addressed to

young children (Won’t Johnny help Mommy?).

17

Example (14) reports how the speaker warned his friend of his upcoming trip. He refers

to himself in the second person, inviting the addressee to imagine herself in his place (he then

continues in the first person to explain that he ended up leaving his friend behind in his village):

(14) ba ya wa go, laa pe e lɛŋ

field put NMLZ in 2SG+3SG told 3SG.REFL to

doo ɓaa zo le Klago

QUOT LOG+COP come PROG K.

‘While in the field, I informed him that I was going to Klango.’

(Literally: “While in the field, you told him that you are going to Klango”.)

[Continued with: “I am going to Klango, but unless I am lying, he is now in Boayaokro.’]

Examples of this kind are not explained by identity issues: there is no question here of

mistaking oneself for another person. It is rather a question of how certain types of potentially

problematic decisions and actions are presented (in this case, the friend was left behind, even

though the speaker claims to have informed him of his upcoming trip). This use of second person

to refer to the actual speaker is a matter of social construction of responsibility, and can hardly be

explained directly by the pronoun’s indexicality.12 Crucially, when a logophoric pronoun appears

in this context, it can hardly be claimed not to refer to the actual speaker: the speaker does not

deny saying these words, and indeed continues reporting the consequences in the first person. The

use of the second person invites the addressee to imagine herself in the same situation, and to

accept the speaker’s implicit justification of his actions.

That both second person and logophoric pronouns refer to the actual speaker poses

problems to lexical accounts of logophoricity, designed to prevent logophoric pronouns from

12 Second person possessive pronouns are also commonly used in Wan to introduce characters

of a story, e.g. ‘There was once your turtle…’. Adding to the typology of “shifted” uses of

second person, Margetts (2015) discusses a number of cases from Saliba-Logea (Oceanic)

where second person pronouns refer to third person participants in narrative contexts.

18

referring to the actual speaker. This restriction applies to cases where the actual speaker is also a

reported one (explaining, for example, the wide-spread restriction on the use of logophoric

pronouns in self-reports, where the reported speaker coincides with the current one).13

3.3 Logophoric pronouns can alternate with first person

A related problem is posed by patterns of alternation between logophoric and first person

pronouns. Nikitina’s (2012a) account predicts that reported speakers should be encoded by

logophoric pronouns if the language has them; similarly, Schlenker’s (2003a) account says

nothing about the possibility of variation in the encoding of non-actual authors. In fact, many

languages offer their speakers a choice between first person and a logophoric pronoun, often

within the same stretch of reported discourse (Boyeldieu 2004a; Nikitina 2012b). In (15), a

logophoric pronoun combines with a co-referential first person pronoun within the same sentence:

a logophoric pronoun is used in a topic position, and a first person pronoun refers back to it from

the position of an oblique argument.

(15) ɓe e ge ee! ɓaa kɛ e,

then 3SG said yes LOG.EMPH that DEF

la nɔni a ŋ mi

2SG lose FUT 1SG at

‘And he said: Yes, as for myself, you won’t be able to recognize me.’

The fact that the reported speaker can be referred to in two different ways within the same

stretch of reported discourse is hard to reconcile with the alleged semantic differences between

first person and logophoric pronouns. Given that the logophoric pronoun is marked as [–actual],

13 This restriction is claimed to be widespread by von Roncador (1988), but exceptions have

nevertheless been reported (Ngbaka, Gokana). If confirmed, the exceptions are potentially

problematic for Schlenker’s (2003a) account.

19

it could only be co-referential with a first person pronoun if the first person pronoun is

underspecified for the same feature, making the use of logophoric pronouns optional. This account

does not explain why both a logophoric and a first person pronoun appear within the same speech

report, and moreover, why they normally appear in a fixed order (a logophoric pronoun is

followed by a co-referential first person pronoun, the reverse order being very rare).

The perspective-based approach does not provide a ready solution. One cannot assume

that parts of the report featuring logophoric pronouns differ constructionally from the parts

featuring first person pronouns. In Wan, at least, neither is more “direct” than the other in any

perceptible way, apart from the occurence of the logophoric pronoun itself. Why a report should

start with a “biperspectival” pronoun and then switch to “monoperspectival” first person remains

unexplained.

4. Proposed tentative solutions

Table 2 summarizes the different issues raised in the previous sections, and in the remaining parts

of the paper I sketch out some tentative solutions.

Table 2. Issues raised by the data in previous sections

Issue problematic for: Summary

logophoric pronouns

appear outside indirect

discourse

Schlenker’s lexical

feature-based account

logophoricity is associated with

indirect discourse; yet in some

languages logophoric pronouns

appear in reports that are otherwise

direct

logophoricity comes in

varieties

Evans’ construction-

oriented account

logophoric languages vary in the way

other indexicals are treated in speech

20

reports, and the difference does not

seem to correspond to a difference in

“directness”

logophoricity need not

involve specialized

pronouns

Evans’ construction-

oriented account

no principled account of the gradient

nature of logophoricity or of the

historical change leading to

specialization of logophoric pronouns

logophoric pronouns are

not restricted to reported

speech or attitude reports

all three accounts some of the logophoric contexts

involve no author, no speaker, and no

reported speech act

logophoric pronouns can

refer to the actual speaker

all three accounts,

especially Schlenker’s

and Nikitina’s lexical

accounts

sometimes the logophoric pronoun

refers to a paticipant who is,

technically, the actual speaker (even

though encoded by second person)

logophoric pronouns can

be co-referential with first

person

all three accounts,

especially Schlenker’s

and Nikitina’s lexical

accounts

first person pronouns may co-refer to

a preceding logophoric pronoun,

sometimes within the same clause

First, we have seen that logophoric pronouns appear outside attitude reports, such as in

aspectual constructions that no longer retain the original meaning of the verb ‘say’. The relevant

construction in Wan involves a control structure: the subject of ‘say’ must be co-referential with

the subject of the following clause. The logophoric pronoun is obligatory on the aspectual reading:

(16) a. yi e ge ɓa kɔ

water DEF say LOG boil

‘The water was about to boil.’ (lit., ‘The water said: let me boil!’)

21

b. # yi e ge ŋ kɔ

water DEF say 1SG boil

The structure associated with the aspectual meaning is more rigid than the one associated

with speech reports. The subject of the logophoric clause must be co-referential with the subject

of the verb ‘say’ (17a), and a loose anaphoric relation is not acceptable (17b).

(17) a. ke yre wati ge ɓa ɓo mɔ

if time say LOG arrive PRT

‘When the time of work comes...’

(Literally, “When the time of work says: let me arrive...”)

b. # ke yre ge ɓa wati ɓo mɔ

if work say LOG time arrive PRT

‘When the time of work comes...’

(Literally, “When the work says: let my time arrive...)

The rigid structure suggests that when the verb ‘say’ is used on the aspectual reading, it

enforces a control relation on the subject of the clause it subcategorizes for.14 This could be

handled by specifying the control relation in the lexical entry of aspectual ‘say’. Since a separate

lexical entry is needed for the aspectual use in any case, it could also be used to specify that the

subject of the embedded clause must be logophoric.

This solution helps define the range of contexts where logophoric pronouns appear.

Logophoric pronouns occurring outside attitude reports no longer seem to be motivated by their

original meaning, just like the use of quotative markers need not be motivated by their original

meaning in all contexts where they are found (Güldemann 2008; see also Matić & Pakendorf

2013). A control-based account of conventionalized uses distinguishes them from uses associated

with speech and attitude reports at the synchronic level.

14 See Nikitina (2008) for more information on control relations in Wan.

22

Second, within speech and attitude reports, the meaning of logophoric pronouns needs to

be revised in light of the actual complexity of the relationship between speech act participant roles

and logophoric pronouns. On the one hand, actual speakers are sometimes referred to by

logophoric pronouns, just as they are also referred to by second and third person pronouns. On

the other hand, non-actual speakers can be referred to by first person pronouns, and such pronouns

may be co-referential with a logophoric pronoun. The two facts suggest that semantic accounts

relying on participant roles, which describe pronominal meaning in terms of notions such as

“speaker”, “author”, or “epistemic validator” (Stirling 1993), are too rigid. They fail to capture

the fluid nature of logophoricity in spoken discourse and patterns of alternation of logophoric and

first person pronouns. A possible solution is to incorporate into the semantic account an additional

discourse dimension describing the participant’s role in the narrative structure, in a way

independent of its role in the speech act.

Crucially, the choice between logophoric and first person pronouns is not determined by

strictly semantic factors but is mediated by categories that operate at the level of discourse

structure. The category that seems most relevant to the data discussed so far is that of narrator.

Incorporating it into the formal account of the meaning of logophoric pronouns helps explain uses

that are problematic on a purely semantic analysis.

The primary function of logophoric pronouns is to distinguish speakers and sources of

attitude reports from the text’s narrator – the agent responsible for interpreting the experience and

turning it into a text. The narrator need not coincide with the actual speaker, since the same story

can be told by different speakers. Actual speakers need not always be constructed as narrators,

since the actual speaker may lend their voice to the story’s characters or claim to be merely

reproducing a text originally produced by someone else. Sometimes an actual speaker may be

disguised as another participant, as in (14), where he refers to himself as an addressee. The

complexities of the relationship between the actual speaker on the one hand and the narrator as

the experience-interpreting agent on the other are reflected in the use of logophoric pronouns.

23

Logophoric pronouns must refer to a speaker or the participant to whom an attitude report

is attributed whenever that participant is not construed as the narrator. First person pronouns, on

the other hand, are not specified, in Wan, for the narrative role: they can refer, within or outside

speech reports, to any speaker or attitude holder, within the limits determined by social and genre-

specific norms of construction of personhood. The underspecified meaning of the first person

explains why logophoric pronouns may alternate with first person pronouns when the actual

speaker represents speech by characters: as long as the characters are distinct from the narrator,

either pronoun can be used. Logophoric pronouns are used in contexts where the actual speaker

claims to reproduce a text by another speaker, acting as a representative of the original narrator:

they appear, for example, in the context of triadic communication when the speaker’s role is

restricted to that of an intermediary (Ameka 2004). Finally, sensitivity to the narrator role explains

why logophoric pronouns appear in contexts where actual speakers represent themselves as

another participant, for example, by referring to themselves as an addressee in (14). Even though

that participant coincides with the actual speaker, it can now be represented as different from the

narrator, creating the special distancing effect we see in that example.

The modified lexical account presents an elegant view of the Wan pronominal system:

all persons are now underspecified for their evaluation context (actual vs. reported). Thus, first

person pronouns can refer either to actual speakers or to story characters; the difference in

interpretation does not imply a syntactic difference in the type of report. As we saw earlier, second

person pronouns, too, can refer to actual or reported addressees; the difference in interpretation

does not imply a syntactic difference or a difference in “perspective”. Different evaluation

contexts may be freely combined within the same report, as we saw in (6).

5. Conclusion

24

This study argued that existing approaches to logophoricity cannot, in their current form,

accommodate discourse data from even one particular language. Lexical accounts seem to do

better than construction-oriented, perspective-based approaches; yet they, too, need to be revised

in light of new data. Several general conclusions can be drawn from this.

First, we still know little about discourse phenomena, and analyses based on elicited data

often have to be revised when natural data becomes available (Nikitina 2018b). Many phenomena

that are currently treated in semantic terms may need to be reconsidered to accommodate

discourse-level categories.

Second, multiple aspects of language structure are subsumed under the notion of

perspective, and even seemingly uniform phenomena such as discourse reporting may not yield

themselves to analyses relying on that notion. Attempts to typologize discourse reporting

strategies based on perspective shifts are unlikely to be productive if European-style perspective

turns out to be irrelevant in some languages. There are indications that entirely different principles

may underly the choice of deictic values in logophoric languages; in Wan, for example, a crucial

role is played by the participant’s role in the narrative structure, rather than the participant’s

involvement in the current speech act. This suggests that the notion of perspective may need

further defining if it is to be used in cross-linguistic comparison.

Third, while the proposed analysis is unconventional in going beyond the standard

semantic and syntactic notions, it relies on notions that are extensively used in narratology and

ethnographic study of language. It accords well with earlier observations of the relevance of

various aspects of narrative structure for a number of morphosyntactic phenomena, ranging from

the use of genre-specific temporal-aspectual forms to the choice of anaphoric elements (Nikitina

2018a). This aspect of linguistic structure has so far remained lamentably understudied, but one

can be hopeful that discourse phenomena – including the notions of perspective and perspective

shift – will receive more sustained scrutiny as the focus of theoretical linguistics shifts to the study

of spoken data from lesser-studied languages.

25

Abbreviations

COP – copula, D – determiner, DEF – definite marker, EMPH – emphatic, FUT – future, M –

masculine, LOG – logophoric, NMLZ – nominalization, PL – plural, POSS – possessor, PROSP –

prospective, PRT – particle, Q – question, QUOT – quotative, REFL – reflexive, SG – singular, SP –

speaker, T – tense

References

Aaron, Uche E. 1992. Reported speech in Obolo narrative discourse. In Shin Ja J. Hwang &

William R. Merrifield (eds.), Language in context: Essays for Robert E. Longacre, 227–

240. Dallas: SIL & The University of Texas at Arlington.

Aikhenvald, Alexandra Y. 2008. Semi-direct speech: Manambu and beyond. Language Sciences

30. 383–422.

Ameka, Felix K. 2004. Grammar and cultural practices: The grammaticalization of triadic

communication in West African languages. Journal of West African Languages 32. 5–

27.

Benveniste, Émile. 1956. La nature des pronoms. Reprinted in Émile Benveniste, Problèmes de

linguistique générale, 251–257. Paris: Gallimard, 1966.

Bhat, D. N. S. 2007. Pronouns. Oxford: Oxford University Press.

Boyeldieu, Pascal. 2004a. A qui s’adresse le logophorique yakoma. In Pascal Boyeldieu & Pierre

Nougayrol (eds.), Langues et cultures: Terrains d’Afrique. Hommage a France Cloarec-

Heiss, 185–191. Leuven: Peeters.

26

Boyeldieu, Pascal. 2004b. Les pronoms logophoriques dans les langues d’Afrique centrale. In D.

Ibriszimow & G. Segerer (eds.), Systèmes de marques personnelles en Afrique, 11–22.

Leuven: Peeters.

Bugaeva, Anna. 2008. Reported discourse and logophoricity in Southern Hokkaido dialects of

Ainu. Gengo Kenky¯u 133. 31–75.

Clements, George N. 1975. The logophoric pronoun in Ewe: Its role in discourse. Journal of West

African Languages 10. 141–177.

Coulmas, Florian. 1986. Reported speech: Some general issues. In Florian Coulmas (ed.), Direct

and indirect speech, 1–28. Berlin: Mouton de Gruyter.

Culy, Christopher. 1994. Aspects of logophoric marking. Linguistics 32. 1055–1094.

Culy, Christopher. 1997. Logophoric pronouns and point of view. Linguistics 35. 845–859.

Dimmendaal, Gerrit J. 2001. Logophoric marking and represented speech in African languages

as evidential hedging strategies. Australian Journal of Linguistics 21. 131–157.

Evans, Nicholas. 2013. Some problems in the typology of quotation: A canonical approach. In

Dunstan Brown, Marina Chumakina & Greville G. Corbett (eds.), Canonical Morphology

and Syntax, 66–98. Oxford: Oxford University Press.

Faltz, Leonard M. 1985. Reflexivization: A study in universal syntax. New York: Garland

Publishing.

Frajzyngier, Zygmunt. 1985. Logophoric systems in Chadic. Journal of African Languages and

Linguistics 7. 23–37.

Frajzyngier, Zygmunt. 1993. A Grammar of Mupun. Berlin: Reimer.

Gentens, Caroline, María Sol Sansiñena, Stef Spronck & An Van Linden. 2019. Irregular

perspective shifts and perspective persistence, discourse-oriented and theoretical

approaches. Pragmatics 29(2). 155–169.

Goddard, Cliff & Anna Wierzbicka. 2019. Direct and indirect speech revisited: Semantic

universals and semantic diversity. Alessandro Capone et al. (eds.), Indirect reports and

pragmatics in the world languages, 173–199. Springer.

27

Goffman, Erving. 1981. Forms of talk. Philadelphia: University of Pennsylvania Press.

Güldemann, Tom. 2008. Quotative indexes in African languages: A synchronic and diachronic

survey. Berlin: Mouton de Gruyter.

Güldemann, Tom & Manfred von Roncador. 2002. Preface. In Tom Güldemann & Manfred von

Roncador (eds.), Reported discourse: A meeting ground for different linguistic domains,

vii–ix. Amsterdam: Benjamins.

Hagège, Claude. 1974. Les pronoms logophoriques. Bulletin de la Société Linguistique de Paris

69(1). 287–310.

Hanks, William F. 1996. Exorcism and the description of participant roles. In Michael Silverstein

& Greg Urban (eds.), Natural histories of discourse, 160–200. Chicago: University of

Chicago Press.

Hellwig, Birgit. 2006. Complement clause type and complementation strategies in Goemai. In R.

M.W. Dixon & Alexandra Y. Aikhenvald (eds.), Complementation, 204–223. Oxford:

Oxford University Press.

Hill, Harriet. 1995. Pronouns and reported speech in Adioukrou. Journal of West African

Languages 25. 87–106.

Hyman, Larry M. 1979. Phonology and noun structure. In Larry M. Hyman (ed.), Aghem

grammatical structure, 1–72. Los Angeles: Department of Linguistics, University of

Southern California.

Hyman, Larry M. & Bernard Comrie. 1981. Logophoric reference in Gokana. Journal of African

Languages and Linguistics 3. 19–37.

Levinson, Stephen C. 1988. Putting linguistics on a proper footing: Explorations in Goffman’s

participation framework. In P. Drew & A. Wootton (eds.), Goffman: Exploring the

interaction order, 161–227. Oxford: Polity Press.

Margetts, Anna. 2015. Person shift at narrative peak. Language 91(4). 755–805.

Matić, Dejan & Brigitte Pakendorf. 2013. Non-canonical ‘say’ in Siberia: Areal and genealogical

patterns. Studies in Language 37(2). 356–412.

28

Nikitina, Tatiana. 2008. The mixing of syntactic properties and language change. Stanford:

Stanford University PhD thesis.

Nikitina, Tatiana. 2012a. Personal deixis and reported discourse: Towards a typology of person

alignment. Linguistic Typology 16(2). 233–263.

Nikitina, Tatiana. 2012b. Logophoric discourse and first person reporting in Wan (West Africa).

Anthropological Linguistics 54(3). 280–301.

Nikitina, Tatiana. 2018a. When linguists and speakers do not agree: The endangered grammar of

verbal art in West Africa. Journal of Linguistic Anthropology 28 (2). 1–24.

Nikitina, Tatiana. 2018b. Focus marking and differential argument marking: The emergence of

bidirectional case marking in Wan. Evangelia Adamou et al. (eds.) Information structure

in lesser-described languages: Studies in prosody and syntax, 195–217. Amsterdam:

John Benjamins.

Nikitina, Tatiana. 2019. Diminutives derived from terms for children: Comparative evidence from

Southeastern Mande. Linguistics 57(1). 1–28.

Nikitina, Tatiana & Anna Bugaeva. Fc. Logophoric speech is not indirect: Towards a syntactic

approach to reported speech constructions. Linguistics.

Nikitina, Tatiana & Yvonne Treis. Fc. The use of manner demonstratives in discourse: A

contrastive study of Wan (Mande) and Kambaata (Cushitic). Åshild Næss, Anna Margetts

& Yvonne Treis (eds.) Demonstratives in Discourse. Berlin: Language Science Press.

Nikitina, Tatiana & Alexandra Vydrina. 2020. Reported speech in Kakabe: Loose syntax with

flexible indexicality. Folia Linguistica.

Perrin, Mona. 1974. Direct and indirect speech in Mambila. Journal of Linguistics 10. 27–37.

Plank, Frans. 1986. Über den Personenwechsel und den anderer deiktischer Kategorien in der

wiedergegebenen Rede. Zeitschrift für germanistische Linguistik 14. 284–308.

Roncador, Manfred von. 1988. Zwischen direkter und indirekter Rede. Tübingen: Niemeyer.

Roncador, Manfred von. 1992. Types of logophoric marking in African languages. Journal of

African Languages and Linguistics 13. 163–182.

29

Rumsey, Alan. 2000. Agency, personhood and the ‘I’ of discourse in the Pacific and beyond.

Journal of the Royal Anthropological Institute 6. 101–115.

Schlenker, Philippe. 2003a. Indexicality, logophoricity, and plural pronouns. In Jacqueline

Lecarme (ed.), Research in Afroasiatic grammar II: Selected papers from the Fifth

Conference on Afroasiatic Languages, 409–428. Amsterdam: John Benjamins.

Schlenker, Philippe. 2003b. A plea for monsters. Linguistics and Philosophy 26. 29–120.

Sells, Peter. 1987. Aspects of logophoricity. Linguistic Inquiry 18. 445–479.

Spronck, Stef & Tatiana Nikitina. 2019. Reported speech forms a dedicated semantic domain.

Linguistic Typology 23(1). 119–159.

Spronck, Stef, An Van linden, Caroline Gentens & María Sol Sansiñena. This issue. Irregular

perspective shift and perspective persistence: What counts as a ‘regular’ perspective

construction?

Stirling, Lesley. 1993. Switch-reference and discourse representation. Cambridge: Cambridge

University Press.

Thomas, Elaine. 1978. A grammatical description of the Engenni language. Dallas: SIL & The

University of Texas at Arlington.

Urban, Greg. 1989. The “I” of discourse. In Benjamin Lee & Greg Urban (eds.), Semiotics, self,

and society, 27–51. New York: Mouton.